Each year in the U.S., one in 8 children is born prematurely and the rate of preterm birth continues to rise. Children born preterm are at substantial risk for language-based learning disabilities that may not be identified until the child enters kindergarten, a delay that precludes opportunities for early intervention. Early behavioral indicators of adverse language and cognitive outcomes are urgently needed, particularly methods that identify underlying deficits amenable to effective interventions. To meet these pressing needs, we have forged an experienced interdisciplinary team with complementary expertise in pediatrics and developmental psycho-linguistics. Our mutual long-term goals are to develop reliable and robust predictors of later outcomes for children born preterm (PT) that allow us to design effective early interventions and to elucidate the neural basis of disorders associated with prematurity and with atypical language development. Recent longitudinal studies of typically-developing infants from the Fernald laboratory have found that online measures of linguistic processing efficiency in very young children predict individual differences in language outcomes several years later. Preliminary results from a joint Feldman-Fernald effort found that 18 to 30 month old PT children had no difficulties participating in the online procedures but that PT children were slower to initiate a response and less successful at maintaining their attention compared to age-matched controls. Therefore, we propose to use these sensitive online measures in a longitudinal study of the development of language processing efficiency in PT children from 18 to 48 mos (n = 70) relative to two SES- and gender-matched comparison groups: (1) full term age-matched children (FT-AM, n = 70) with the full range of language abilities, and (2) full term language- matched children (FT-LM, n = 70) who are matched individually to the PT children on age and language level. Children will be tested longitudinally at 18, 24, 30, and 36 mos on a series of increasingly challenging, but developmentally appropriate, assessments of online language comprehension and on standardized offline tests of cognition and language. We hypothesize that PT children will perform more poorly than both groups of FT peers, and that individual differences in online processing efficiency will differentiate those children with good outcomes from those who demonstrate persistent language or associated difficulties in late preschool. We also hypothesize that clinical and/or neurological factors, such as severity of white matter injury, will correlate with language and cognitive outcomes. However, those relations will be reduced after controlling for linguistic processing efficiency, suggesting that speed and accuracy of language processing mediate the link between the neuropathology of prematurity and poor language functioning in PT children. This innovative translational project adopts a developmental psycholinguistic approach to explain individual variation in outcomes in a clinical population at high risk for language disorders, and is poised to inform future theory- and empirically-driven intervention research as well as studies of the neural basis of language disorders.